Early Prediction of Fatigue Induced

Nucleation of Microcracks by Magnetic Barkhausen Noise

Cesar BUQUE, RTD Group, Rotterdam, The Netherlands

Abstract. A new method of producing and monitoring magnetic Barkhausen noise (BN) has been developed and applied to scan for locations of future microcrack nucleations caused by fatigue on the surface of ferromagnetic materials. This paper deals with the results of the preliminary investigations, which for the purpose of understanding the physical background, were focused on nickel crystals. Massive nickel probes were fatigued gradually at constant conditions (Temperature and plastic strain amplitude) until the occurrence of a fatigue failure. The fatigue experiments were stopped several times for scanning the specimen surface. The evaluation of the spatial variation of the integral BN leads to the conclusion that macroscopic compressive internal stresses exist in the so-called persistent slip bands (PSBs) which consist of a typical dislocation configuration that is considered precursor of fatigue cracks. The results showed that BN is a powerful method for a very early prediction of micro-crack locations.

1. Introduction

Ferritic stainless steels have been increasingly used in elevated temperature applications because of their economical combinations of mechanical and corrosion properties. They are currently considered to be prime structural materials for first wall and blanket of future thermonuclear fusion reactors. For these applications, structural components are subjected to repeated thermal stresses as a result of the burning operation of fusion reactors. Such a repeated thermal load can lead to fatigue of the structural materials. Despite the importance acquired by these steels in the last years, only a few works have been found in the literature about their fatigue behavior. Fatigue failure is the outcome of gradual weakening of the material. It involves a hierarchy of micro-mechanisms and processes ranging from atomic dimensions to typical “macroscopic” dimensions, such as grain size in the case of a polycrystalline material. In a nominally defect-free, high purity material subjected to fluctuating loads, the progression of fatigue damage generally occurs in the following stages: (i) Nucleation of microscopic defects and creation of irreversible changes in the microstructural and substructural constitution of the material which are accompanied by cyclic hardening and softening. (ii) Formation of a dominant macroscopic flaw by multiplication and annihilation of the microscopic defects. (iii) Subcritical formation and growth of microcracks. (iv) Final failure of the material or instability.

ECNDT 2006 - Tu.1.6.1

1

A good knowledge of the behavior of the materials under fatigue is necessary for calculation of the stress-strain fields and for the life-time prediction of structural elements. The more precise the knowledge on the fatigue caused material changes the more accurate is the life time prediction. The best methods should be able to allocate and monitor the regions for future nucleation of microcracks, very far before they nucleate, that means in the stages (i) and (ii) as described above. Several methods have been proposed to achieve that knowledge. Most of them are destructive. Ferritic steels however are ferromagnetic materials and this property can be used to access their mechanical behavior in a non-destructive manner. For example, by the use of the magnetic barkhausen noise (BN) method. The aim of the present work is to analyze in a very first step, how far the BN method can be applied for the assessment of very localized material property variations, caused by fatigue and leading to a later formation of microcracks. In these preliminary studies the studies were made on nickel single and polycrystals, which can be considered model materials with ferromagnetic properties. 2. Physical background 1.1 Fatigue behaviour of fcc metals

Fatigue of fcc metals leads to the evolution of spatially heterogeneous dislocation distributions on different scales. If the plastic load exceeds a certain level, macroscopically extended persistent slip bands (PSBs) appear in so called glide active directions. PSBs represent material zones with high strain localization and most frequently they act as preferential sites for microcrack nucleation (Fig. 1). The PSBs are characterised by a ladder-like dislocation patterns (Fig. 2), which are embedded in a matrix with a bundle (veins) dislocation configuration [1 - 6].

Fig. 1: SEM micrographs illustrating the role of persistent slip bands as crack initiators. The images were taken in a sharp-corner prepared nickel single crystal specimen. The extrusions caused by the highly localised deformation in the persistent slip bands are clear recognisable. As a consequence microcracks are formed in the interfaces between PSB and matrix. Note that such extrusions can only be seen in an electron microscope and even then, only if the PSBs are oriented in a strong favourable manner (slip system orientation) to the surface. A NDT method which is able to detect these microcrack locations far before they nucleate must be able to detect PSB locations.

2

Fig. 2: Non-destructive image of dislocation structure in a nickel specimen. The ladder-like dislocation arrangements of dislocation structures characteristic of PSBs are clear recognisable. Note the heterogeneous distribution of PSBs and that PSBs appear isolated or in groups forming so called macrobands.

The Fig. 2 shows dislocations structures stemming from a fatigued specimen. The picture was taken using a non-destructive way of imaging the dislocation structures in a scanning electron microscope, electron channelling contrast. Internal stresses arise owing to the heterogeneous plastic associated with the non-homogeneous dislocation distribution. According to the special slip model being considered the bands may be in tension or in compression. In the present work, along the loading axis variations of internal stresses of fatigued nickel crystals are studied by local measurements of magnetic Barkhausen Noise.

1. 2. Parameters describing the Barkhausen noise signal

It is well known that internal and/or applied stresses influence the Barkhausen noise which is generated by jerky changes in the magnetization of a ferromagnetic material under an applied alternating magnetic field. The Barkhausen noise has high sensitivity to such internal/applied stresses, therefore, detecting and evaluating the BN provides a suitable tool for analysing the stress in fatigued ferromagnetic crystals. In doing so, it should be possible to detect the locations with high stress concentrations, which are the PSB locations.

Different parameters can be used to describe and evaluate the BN. In the simplest case the mean BN level or integrated BN (iBN) during the magnetization cycle is determined [7 - 12]. In a more informative method the distribution of the BN magnitude along the whole magnetization half-loop, i.e. the enveloping curve on the BN (ecBN), may be detected. For nickel it has been shown that at room temperature the energy contribution of the material anisotropy to the magnetization energy becomes very small, so that already low applied or internal stresses determine the favourable directions of magnetization (Fig. 3). Considering the negative magnetostriction constant of nickel, the direction of magnetization in a magnetic

3

0 MPa

-5 MPa 5MPa

-12MPa 12MPa

-25MPa 25MPa

Fig. 3: Influence of stresses on the total magnetic energy of nickel materials. The total magnetic energy considered here consists of magnetostrictive energy and crystallographic energy.

4

domain will be essentially parallel to the direction of the compressive stress and perpendicular to the direction of the tensile stress in the coordinate system of the principle axis (Fig. 3). The response of the spontaneous jerks to the magnetization during a magnetizing cycle is mostly measured in the direction (measuring direction MD) of the external magnetic field which e loading axis (LA). If the stress state changes from a tensile to a compressive state, the integral BN and the distribution of the BN magnitude in the half-loop are expected to vary in very typical way for nickel at room temperature. The iBN should increase with decreasing tensile stress and increasing compressive stress because the direction of the spontaneous magnetization in the domains tends more and more towards the loading axis LA and, with that, towards the measuring direction MD. Thus, the component of magnetization along the measuring direction will be increased, which means that magnetization changes will produce larger noise signals. Beside this, domain wall jumping processes become more favourable in comparison with a noiseless mere turning of the magnetization in the magnetic domains. With increasing compressive stresses the BN magnitude distribution ecBN should shift along the magnetization loop from lower external magnetic fields to larger ones, owing to the fact during magnetization the nucleation of domains with a magnetization parallel to LA and to H is supported by compressive stresses. Thus, starting the magnetization processes at the maximum external magnetic field, the larger the magnitude of the compressive stress, the sooner the wall movement will begin. Recently it have been shown that during the elastic deformation of a ferromagnetic specimen the mean BN level as well as the enveloping curve ecBN depend on the applied stress in the expected way. In the case of nickel the iBN is oserved to decrease with decreasing compressive stress and increasing tensile stress. For iron, which is one of the main components in ferritic alloys, the reverse stress influence was found, reflecting that the opposite magnetostriction of nickel and iron has the opposite sign. As excepted, starting from the saturation of the magnetization, the BN in nickel sets in at lower external magnetic fields when the compressive stress decreases and the tensile stress increases. The results presented in this paper were obtained with an innovative technique for producing and monitoring BN. The technique which allows detection of the integral signal BN as well as the BN enveloping curve at different sites on the specimen surface.

1.3 Experimental details

Nickel single and polycrystals were fatigued at room and at high temperature. The fatigue tests were mainly performed at medium amplitudes of the plastic strain amplitude. At such amplitudes PSBs are expected to occupy a certain volume of the fatigued material. The geometry of the probes and the apparatus for generating and measuring the BN are shown in Fig. 4. The frequency of a.c. powersupply was 50 Hz. As can be seen the applied magnetic field is oriented parallel to the specimen axis LA. For detecting the local BN response, the special scanner sensor (which has a configuration very similar to that of a sound head) is arranged above the specimen. This means that the magnetic measuring direction MD lies parallel to the direction of the external magnetic field H. The sensors consist of a special ferritic alloy with a small slit of 4 µm and a slit length varying from 0.20 mm and 0.80 mm. The electronic equipment enables us to perform two evaluation procedures. On the one hand, the integral BN voltage (r.m.s. Ui), resulting from all the magnetization jerks along the hysteresis loop. The results can be viewed on a display as a function of the probe displacement in a three-dimensional plot (local iBN surface distribution). In spite of the small slit width of the magnetic sensor the iBN signal at a given surface position is detected with a resolution of 50 to 100 µm. on the other hand, it is

5

possible to measure the enveloping curve of the BN at fixed places on the specimen surface. The ecBN is obtained as indicated schematically Fig. 4. A small magnetic field interval (window) is defined and can be shifted along the whole magnetization curve. By determining the BN r.m.s. voltage Uw in the curve at different sites, a BN signal can be detected as a function of the applied field H and thus, the distribution of the BN magnitude along the hysteresis loop is measured. Similar technology was patented and is being applied by the Fraunhofer Gesellschaft Saarbrücken [13-14], mainly for the determination of surface stress distributions (internal stresses).

a) c)

b)

d)

Fig. 4: Schematic illustration of the specimen geometry used for the fatigue tests (a); the experimental arrangement for producing and scanning the BN (b); a barkhausen noise graph (c); and a magnetic hysteresis loop illustrating the influence of magnetization jerks on the hysteresis behaviour (d).

2. Results

The evolution of fatigue behaviour is related to the surface distribution of the iBN on the surface as shown in fig. 3. With the appearance of PSBs a marked heterogeneity of the iBN surface distribution is clearly recognisable. The high iBN zones belong to surface regions with a high density of PSBs (macroband regions). Small iBN levels were found only on locations without PSBs. Taking into account the low spatial resolution of the local BN measurement is becomes clear that a direct correlation between the iBN distribution and the positioning of an individual PSB or a group of only a small number of PSBs cannot de

6

expected. Further fatigue of the specimen result in a more pronounced heterogeneity of the local distribution of the iBN. More recent measurements on very coarse grained material show also that the technology principle is also applicable for polycrystals. The results are summarised in the figures below.

Fig. 5: Distribution of the iBN voltage Ui on the surface of a single nickel specimen. With increasing number of fatigue cycles (from top to bottom) there is first an increase of the iBN voltage in PSB locations and after a certain cyclic number the magnitude of the iBN voltage in the PSB locations decreases drastically. This effect coincides exactly with the nucleation moment of micro cracks. The internal stresses that were concentrated at the PSB location sites relax with the appearance of PSBs.

7

Fig. 6: Evolution of surface and material damage on polycrystalline nickel due to fatigue. The bright straight traces in the images correspond to PSBs. In the last picture, which corresponds to a fatigue stage just before the fatigue failure, the appearance of microcracks related to PSBs is clear recognisable.

8

Fig. 5: Typical distribution of the iBN voltage on the surface of a grain of a coarse grained nickel polycrystalline specimen. The performance of the measurement results regarding PSB distribution and other fatigue related properties is very difficult in the case of polycrystalline materials. This is most probably due to anisotropic properties. For crystallographic orientation of the grains must be known in order to get the optimum results, which is not so realistic for field applications. For very large grained materials however, the scanning of the surface ferromagnetic barkhausenn noise is in laboratory scale a good method for the early detection of microcracrack nucleation sites.

3. Conclusions:

Locations with PSBs can be pretty well detected by the magnetic barkhausen noise method. Inhomogeneities in the iBN signal are correlated with existence of PSBs. The high iBN

9

levels above PSB surface regions are due to mean compressive stresses in the PSB macrobands. Once microcracks nucleate the magnitude of the iBN signal drops and this is interpreted in terms of the relaxation of internal stresses due to the micro cracking. Polycrystalline materials can also be investigated by such a scanning method if the underlying microstructure (grains) are very large. From these very preliminary studies it can be concluded that the magnetic barkhausen noise is a powerful method for the early detection of micro crack locations. The scanning method has strong limitations in the case of polycrystalline materials. The anisotropy and the stress constraints caused by the differently oriented grains must be considered for the right interpretation of the signal. To use the magnetic barkhausen noise methos for the early detection of microcrack nucleation other technology rather than scanning the surface must be applied. This will be discussed during the ECNDT and interested readers are directed to the literature given at end of this article. A very interesting application of this technology was developed by the Fraunhofer Gesellschaft in Saarbrücken [13-14]. In combination with eddy current technology Altpeter et al [13-14] use a similar technology for the determination os surface stress distributions in different ferromagnetic materials.

References

[1] U. Holzwarth and U. Essmann, Appl. Phys. A 57 (1993), pp. 131–141. [2] H. Mughrabi, Phys. Status Solidi (a) 104 (1987), pp. 107–120. [3] C. Holste and H.-J. Burmeister, Phys. Status Solidi (a) 57 (1980),

pp. 269–280. [4] U. Essmann, U. Gösele and H. Mughrabi, Philos. Mag. A44 (1981), pp.

405–426. [5] L.M. Brown In: M.F. Holby, R. Bullonth, C.S. Hartley and J.P.

Hirth, Editors, Dislocation Modelling of Physical Systems, Pergamon, Madrid (1981), pp. 51–68. [6] C. Blochwitz and E. Kahle, Krist. Technol. 15 (1980), pp. 977–986. [7] H.-D. Schumann, Phys. Status Solidi (a) 18 (1973), pp. k27–k29. [8] W. Willmann and G. Wollmann, Exp. Technol. Phys. 31 (1983), pp. 533–543. [9] W. Tirschler and H.-D. Schumann, Proc. DGZfP DachJahrestagung 1991 Zerstörungsfreie Materialprüfung Luzern (1991), pp. 682–689. [10] E. Kneller, Ferromagnetism, Springer, Cambridge (1962). [11] C. Buque and W. Tirschler, Wiss Z. Tech. Univ. Dresden 43 (1993), pp. 33–36. [12] C. Buque, W. Tirschler and C. Blochwitz, Z. Metall. 86 (1995), pp. 671–681. [13] Altpeter, I.; Hoffmann, J.; Kopp, M.; and Grimm, H. Prak. Metallogr. 37 (2000) 5. [14] Altpeter, I. and Meyendorf, N. in Heft 7/89 (1989) Galvaotechnik.

10